Question_Answering.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/ajits-github/NLP_with_Hugging_Face_Transformers/blob/main/Question_Answering.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "9f7992af",
      "metadata": {
        "id": "9f7992af",
        "outputId": "8602561c-f1ad-4580-8e48-0ffab7a71c71"
      },
      "outputs": [
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "/opt/anaconda3/envs/transformers-2/lib/python3.9/site-packages/tqdm/auto.py:22: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
            "  from .autonotebook import tqdm as notebook_tqdm\n"
          ]
        }
      ],
      "source": [
        "import transformers\n",
        "\n",
        "#Set to avoid warning messages.\n",
        "transformers.logging.set_verbosity_error()"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "848467ee",
      "metadata": {
        "id": "848467ee"
      },
      "source": [
        "## 01.03. Using a Qu-An Pipeline"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "ac6163bf",
      "metadata": {
        "id": "ac6163bf",
        "outputId": "e6b9d238-7778-407b-dc5c-95d9f3a0e1b7"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "{'score': 0.9553402066230774, 'start': 327, 'end': 330, 'answer': '29%'}\n"
          ]
        }
      ],
      "source": [
        "from transformers import pipeline\n",
        "\n",
        "context=\"\"\"\n",
        "Earth is the third planet from the Sun and the only astronomical object\n",
        "known to harbor life. While large volumes of water can be found\n",
        "throughout the Solar System, only Earth sustains liquid surface water.\n",
        "About 71% of Earth's surface is made up of the ocean, dwarfing\n",
        "Earth's polar ice, lakes, and rivers. The remaining 29% of Earth's\n",
        "surface is land, consisting of continents and islands.\n",
        "Earth's surface layer is formed of several slowly moving tectonic plates,\n",
        "interacting to produce mountain ranges, volcanoes, and earthquakes.\n",
        "Earth's liquid outer core generates the magnetic field that shapes Earth's\n",
        "magnetosphere, deflecting destructive solar winds.\n",
        "\"\"\"\n",
        "\n",
        "quan_pipeline = pipeline(\"question-answering\",\n",
        "                         model=\"deepset/minilm-uncased-squad2\")\n",
        "\n",
        "answer=quan_pipeline(question=\"How much of earth is land?\",\n",
        "             context=context)\n",
        "print(answer)\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "bb71eb82",
      "metadata": {
        "id": "bb71eb82",
        "outputId": "149f645d-2d12-4ed1-94be-c57027650264"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\n",
            "Another question :\n",
            "{'score': 0.2615407705307007, 'start': 399, 'end': 471, 'answer': \"Earth's surface layer is formed of several slowly moving tectonic plates\"}\n"
          ]
        }
      ],
      "source": [
        "print(\"\\nAnother question :\")\n",
        "print(quan_pipeline( question=\"How are mountain ranges created?\",\n",
        "             context=context))"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "c7999e87",
      "metadata": {
        "id": "c7999e87"
      },
      "source": [
        "## 01.05 Evaluating Qu-An Performance"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "c11d8e01",
      "metadata": {
        "id": "c11d8e01",
        "outputId": "56a68fb7-59ac-4a19-8bae-631c6dfd6f01"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "F1 is 100.0  for answer : Paris\n",
            "F1 is 0.0  for answer : London\n",
            "F1 is 22.22222222222222  for answer : Paris is one of the best cities in the world\n",
            "\n",
            " Cumulative Results : \n",
            " {'exact': 33.333333333333336, 'f1': 40.74074074074074, 'total': 3, 'HasAns_exact': 33.333333333333336, 'HasAns_f1': 40.74074074074074, 'HasAns_total': 3, 'best_exact': 33.333333333333336, 'best_exact_thresh': 0.0, 'best_f1': 40.74074074074074, 'best_f1_thresh': 0.0}\n"
          ]
        }
      ],
      "source": [
        "from evaluate import load\n",
        "squad_metric = load(\"squad_v2\")\n",
        "\n",
        "#Ignoring Context & Question as they are not needed for evaluation\n",
        "#This example is to showcase how the evaluation works based on match between the prediction\n",
        "#and the correct answer\n",
        "\n",
        "correct_answer=\"Paris\"\n",
        "\n",
        "predicted_answers=[\"Paris\",\n",
        "                 \"London\",\n",
        "                 \"Paris is one of the best cities in the world\"]\n",
        "\n",
        "cum_predictions=[]\n",
        "cum_references=[]\n",
        "for i in range(len(predicted_answers)):\n",
        "\n",
        "    #Use the input format for predictions\n",
        "    predictions = [{'prediction_text':predicted_answers[i],\n",
        "                    'id': str(i),\n",
        "                    'no_answer_probability': 0.}]\n",
        "    cum_predictions.append(predictions[0])\n",
        "\n",
        "    #Use the input format for naswers\n",
        "    references = [{'answers': {'answer_start': [1],\n",
        "                               'text': [correct_answer]},\n",
        "                   'id': str(i)}]\n",
        "    cum_references.append(references[0])\n",
        "\n",
        "    results = squad_metric.compute(predictions=predictions,\n",
        "                                   references=references)\n",
        "    print(\"F1 is\", results.get('f1'),\n",
        "          \" for answer :\", predicted_answers[i])\n",
        "\n",
        "#Compute for cumulative Results\n",
        "cum_results=squad_metric.compute(predictions=cum_predictions,\n",
        "                                 references=cum_references)\n",
        "print(\"\\n Cumulative Results : \\n\",cum_results)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "e55d5988",
      "metadata": {
        "id": "e55d5988"
      },
      "outputs": [],
      "source": []
    }
  ],
  "metadata": {
    "kernelspec": {
      "display_name": "Python 3 (ipykernel)",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.9.15"
    },
    "colab": {
      "provenance": [],
      "include_colab_link": true
    }
  },
  "nbformat": 4,
  "nbformat_minor": 5
}